[PATCH v3 4/7] kunit: Handle test faults

[Mickaël Salaün]

On Sat, Mar 23, 2024 at 03:37:21PM +0800, David Gow wrote:
> On Tue, 19 Mar 2024 at 18:49, Mickaël Salaün <mic at digikod.net> wrote:
> >
> > Previously, when a kernel test thread crashed (e.g. NULL pointer
> > dereference, general protection fault), the KUnit test hanged for 30
> > seconds and exited with a timeout error.
> >
> > Fix this issue by waiting on task_struct->vfork_done instead of the
> > custom kunit_try_catch.try_completion, and track the execution state by
> > initially setting try_result with -EINTR and only setting it to 0 if
> > the test passed.
> >
> > Fix kunit_generic_run_threadfn_adapter() signature by returning 0
> > instead of calling kthread_complete_and_exit().  Because thread's exit
> > code is never checked, always set it to 0 to make it clear.
> >
> > Fix the -EINTR error message, which couldn't be reached until now.
> >
> > This is tested with a following patch.
> >
> > Cc: Brendan Higgins <brendanhiggins at google.com>
> > Cc: Shuah Khan <skhan at linuxfoundation.org>
> > Reviewed-by: Kees Cook <keescook at chromium.org>
> > Reviewed-by: David Gow <davidgow at google.com>
> > Tested-by: Rae Moar <rmoar at google.com>
> > Signed-off-by: Mickaël Salaün <mic at digikod.net>
> > Link: https://lore.kernel.org/r/20240319104857.70783-5-mic@digikod.net
> > ---
> >
> > Changes since v2:
> > * s/-EFAULT/-EINTR/ in commit message as spotted by Rae.
> > * Add a comment explaining vfork_done as suggested by David.
> > * Add David's Reviewed-by.
> > * Add Rae's Tested-by.
> >
> > Changes since v1:
> > * Add Kees's Reviewed-by.
> > ---
> >  include/kunit/try-catch.h |  3 ---
> >  lib/kunit/try-catch.c     | 19 ++++++++++++-------
> >  2 files changed, 12 insertions(+), 10 deletions(-)
> >
> > diff --git a/include/kunit/try-catch.h b/include/kunit/try-catch.h
> > index c507dd43119d..7c966a1adbd3 100644
> > --- a/include/kunit/try-catch.h
> > +++ b/include/kunit/try-catch.h
> > @@ -14,13 +14,11 @@
> >
> >  typedef void (*kunit_try_catch_func_t)(void *);
> >
> > -struct completion;
> >  struct kunit;
> >
> >  /**
> >   * struct kunit_try_catch - provides a generic way to run code which might fail.
> >   * @test: The test case that is currently being executed.
> > - * @try_completion: Completion that the control thread waits on while test runs.
> >   * @try_result: Contains any errno obtained while running test case.
> >   * @try: The function, the test case, to attempt to run.
> >   * @catch: The function called if @try bails out.
> > @@ -46,7 +44,6 @@ struct kunit;
> >  struct kunit_try_catch {
> >         /* private: internal use only. */
> >         struct kunit *test;
> > -       struct completion *try_completion;
> >         int try_result;
> >         kunit_try_catch_func_t try;
> >         kunit_try_catch_func_t catch;
> > diff --git a/lib/kunit/try-catch.c b/lib/kunit/try-catch.c
> > index cab8b24b5d5a..7a3910dd78a6 100644
> > --- a/lib/kunit/try-catch.c
> > +++ b/lib/kunit/try-catch.c
> > @@ -18,7 +18,7 @@
> >  void __noreturn kunit_try_catch_throw(struct kunit_try_catch *try_catch)
> >  {
> >         try_catch->try_result = -EFAULT;
> > -       kthread_complete_and_exit(try_catch->try_completion, -EFAULT);
> > +       kthread_exit(0);
> 
> It turns out kthread_exit() is not exported, so this doesn't work if
> KUnit is built as a module.
> 
> I think the options we have are:
> - Add EXPORT_SYMBOL(kthread_exit).

I'll go this way. This should not be an issue because
kthread_complete_and_exit() can already (only) call kthread_exit() if
the completion is NULL, but directly calling kthread_exit() is cleaner.

> - Keep using out own completion, and kthread_complete_and_exit()
> - try_get_module() before spawning the thread, and use
> module_put_and_kthread_exit().
> 
> I think all of these would be okay, but I could've missed something.
> 
> -- David